Another day, another graph funding round. Last week it was TigerGraph, this week it's Katana Graph. Katana Graph, a high-performance scale-out graph processing, AI, and analytics company, announced it has completed a $28.5 million Series A financing round led by Intel Capital.
If you are wondering what is that about and where did it come from -- good questions. ZDNet connected with Keshav Pingali, Katana Graph CEO and co-founder, to get the answers.
The origin of Katana Graph
Pingali has a background in programming parallel systems, runtime systems, and compilers, having held professorships at Cornell and the University of Texas at Austin, where he leads a research group. Chris Rossbach, Katana Graph CTO and co-founder, is also an assistant professor at UT Austin.
Rossbach has also held senior roles at VMware and Microsoft Research. Katana's board also includes Farshid Sabet, chief business officer, who was GM of Edge AI at Intel, and Amy Chang, board member at Procter and Gamble and Cisco, as a board advisor.
Katana was not exactly spun out of a campus by graduate whiz kids. Pingali's long line of research was funded by several DARPA projects, and one of those DARPA projects was led by BAE. They were building a system for real=time intrusion detection in computer networks and wanted to mine very large interaction graphs to reveal forbidden patterns.
A number of graph database vendors were trialed, but none met the project's requirements. Pingali and Rossbach were contacted and were able to build a solution that worked. Ironically, the solution was never deployed as the project did not fly, but the wheel was set in motion.
The solution Pingali and Rossbach built was also trialed in different projects, such as chip design and mesh generation, and passed the test with flying colors. Pingali and Rossbach were encouraged to start a company around their "very versatile, very fast, very scalable graph engine."
Pingali referred to use cases where they were asked to ingest and process graphs having 4.3 trillion edges. In those cases, he went on to add, both ingestion speed and analytics were an issue, but Katana was able to deal with both.
The Ferrari of Graph?
This provided some insight as to the origins and nature of Katana Graph. It was not entirely clear to us whether Katana was an analytics engine, or a fully-fledged graph database. Analytics engines serve analytics use cases, while databases are more versatile.
Pingali noted that Katana offers graph database and graph analytics functionality, plus a Graph API. The Katana team started out focused on analytics, and built a database to go with that a bit later. That was after they had tried to use existing databases to accommodate the analytics engine and found them not up to the task at their scale, Pingali said:
"We started from the analytics engine, the compute engine, and then realized that we need an integrated storage solution as well. Then on top of that, it's relatively straightforward for us to build these high-performance, AI, and machine learning libraries."
That made us wonder whether Katana Graph leverages custom hardware. Pingali clarified that Katana supports heterogeneous clusters of computing resources including x86 CPUs, ARM CPUs, GPUs, and other accelerators. They will expand to more accelerators, he went on to add, depending on customer demand.
Speaking of customers, we also wondered about the kind of customer that needs to ingest and do analytics on a 4.3 trillion edge graph. Pingali confirmed the obvious -- that would be organizations of the massive kind. He went on to add, however, that Katana is not only cut out for those:
"We have a competitive advantage, even if the graphs are relatively small and fit on a single machine. I don't want to leave the impression that we're like the Ferrari of this ecosystem, and it's only if you can drive it to 250 miles an hour down the highway that Katana becomes useful. We have seen competitive advantages across the board in terms of end to end performance."
Use cases and growth
Katana's Graph API can be accessed in both C++ and Python, specifically addressing data scientists. Katana also comes with support for Jupyter Notebooks, as well as many algorithms and vertical-specific customizations out of the box. But a query language does exist too.
Katana supports the property graph data model, and openCypher, the open-source query language originally contributed by Neo4j. Using a popular query language rather than coming up with their own was something the people at Katana were eager to do, and also prompted by customers to adopt.
The Series A financing builds on what Pingali said was an exceptional year for Katana Graph, which saw a rapidly-growing number of enterprise clients in the pharmaceutical, fintech, identity, security, and Electronic Design Automation market segments, as well as strong momentum in the big data analytics market.
Besides Intel Capital, existing and new investors including Walden International (WRVI Capital), Nepenthe Capital, Dell Technologies Capital, and Redline Capital Management participated in the round.
Intel Capital investment director, Vijay Reddy, will join Katana Graph's board of directors, while Redline's general partner Tatiana Evtushenkova, and Dell Technologies Capital president, Scott Darling, will join as board observers.
That sounds like enough business experience to go by, and Pingali painted a very promising picture for Katana's growth. He noted that they have a long queue of customers waiting to be served, and the funding will enable Katana to grow its current headcount of 30 to serve as many customers as possible, without losing focus.
Not bad at all, for a company that got its seed funding in May 2020. Katana Graph seems to be cutting it indeed.